Selected work

Representative Papers

Calibrated Similarity for Reliable Geometric Analysis of Embedding Spaces

Jan 23, 2026

This work addresses the limited interpretability of cosine similarity in pretrained embedding spaces, where absolute similarity values are concentrated in a narrow range due to anisotropy. The authors propose a monotonic calibration method based on isotonic regression that reparameterizes similarity scores without altering the underlying embeddings or their geometric structure. The approach strictly preserves the original ranking order of similarities, thereby maintaining all ordinal-dependent structures—such as nearest neighbors, angular rankings, and threshold graphs—intact. The calibrated similarities achieve near-perfect alignment in absolute values while retaining 98% local stability under seven types of perturbations and fully preserving rank correlation with the original similarities.

1 citations1 influentialRead paper

ZAPS: Zero-Cost Active Proxy Search for Neural Architecture Search

Sep 12, 2026

ZAPS通过结合零成本代理和架构拓扑，利用主动学习框架解决神经架构搜索中评估耗时问题。

0 citationsRead paper

On the Complexity of Entrywise Power Matrix Factorization

Jul 06, 2026

This study investigates the computational complexity of element-wise power matrix factorization (EPMF): given a nonnegative matrix \(X\), a rank \(r\), and a real exponent \(p\), the task is to determine whether there exists a rank-\(r\) matrix \(X_r\) such that \(X = |X_r|^{\circ p}\) exactly, or approximately in Frobenius norm. By reducing exact EPMF to the sign-flipping problem, the authors establish for the first time that exact EPMF is strongly NP-hard—previously only weak NP-hardness was known—and further show that it becomes polynomial-time solvable when the rank is fixed and fixed-parameter tractable when parameterized by the input rank. For the approximate variant, they prove NP-hardness even when \(r = 2\). This work thus provides a complete characterization of the complexity landscape of EPMF in both exact and approximate settings.

0 citationsRead paper

A Decision-Making Framework for New Member Integration in Renewable Energy Communities under Prospect Theory

Jun 29, 2026

This study addresses the challenge of coordinating long-term investment pricing with short-term energy scheduling during the integration of new members into renewable energy communities. To this end, the authors propose a bi-level framework that combines extensive-form games with generalized Nash equilibrium to separately model strategic long-term decisions and day-ahead operational scheduling. Notably, prospect theory is incorporated for the first time in this context to capture heterogeneous user preferences and bounded rationality. Through multi-scenario numerical simulations on a community comprising five existing members and eleven candidate users, the proposed approach demonstrates superior performance over existing heuristic metrics and quantitatively reveals the critical influence of reference point selection, decision sequence, and behavioral parameters on system equilibrium outcomes.

0 citationsRead paper

Manifold-based Algorithms for the Hadamard Decomposition

May 27, 2026

This work addresses the Hadamard decomposition problem—recovering two low-rank matrices whose element-wise product approximates a given matrix. By reformulating the problem as a structured matrix factorization with explicit constraints, the authors propose three efficient algorithms: a direct decomposition method based on Manopt, a block projected gradient approach, and a projection-free Riemannian gradient descent scheme. A novel initialization strategy is also introduced to enhance solution accuracy. The proposed methods are particularly well-suited for large-scale sparse data and demonstrate significant improvements over truncated SVD and existing Hadamard decomposition techniques on both synthetic and real-world datasets, confirming their computational efficiency and competitive performance.

0 citationsRead paper

Recent publications

Latest Papers

ZAPS: Zero-Cost Active Proxy Search for Neural Architecture Search

Sep 12, 2026

ZAPS通过结合零成本代理和架构拓扑，利用主动学习框架解决神经架构搜索中评估耗时问题。

0 citationsRead paper

On the Complexity of Entrywise Power Matrix Factorization

Jul 06, 2026

This study investigates the computational complexity of element-wise power matrix factorization (EPMF): given a nonnegative matrix \(X\), a rank \(r\), and a real exponent \(p\), the task is to determine whether there exists a rank-\(r\) matrix \(X_r\) such that \(X = |X_r|^{\circ p}\) exactly, or approximately in Frobenius norm. By reducing exact EPMF to the sign-flipping problem, the authors establish for the first time that exact EPMF is strongly NP-hard—previously only weak NP-hardness was known—and further show that it becomes polynomial-time solvable when the rank is fixed and fixed-parameter tractable when parameterized by the input rank. For the approximate variant, they prove NP-hardness even when \(r = 2\). This work thus provides a complete characterization of the complexity landscape of EPMF in both exact and approximate settings.

0 citationsRead paper

A Decision-Making Framework for New Member Integration in Renewable Energy Communities under Prospect Theory

Jun 29, 2026

This study addresses the challenge of coordinating long-term investment pricing with short-term energy scheduling during the integration of new members into renewable energy communities. To this end, the authors propose a bi-level framework that combines extensive-form games with generalized Nash equilibrium to separately model strategic long-term decisions and day-ahead operational scheduling. Notably, prospect theory is incorporated for the first time in this context to capture heterogeneous user preferences and bounded rationality. Through multi-scenario numerical simulations on a community comprising five existing members and eleven candidate users, the proposed approach demonstrates superior performance over existing heuristic metrics and quantitatively reveals the critical influence of reference point selection, decision sequence, and behavioral parameters on system equilibrium outcomes.

0 citationsRead paper

Manifold-based Algorithms for the Hadamard Decomposition

May 27, 2026

This work addresses the Hadamard decomposition problem—recovering two low-rank matrices whose element-wise product approximates a given matrix. By reformulating the problem as a structured matrix factorization with explicit constraints, the authors propose three efficient algorithms: a direct decomposition method based on Manopt, a block projected gradient approach, and a projection-free Riemannian gradient descent scheme. A novel initialization strategy is also introduced to enhance solution accuracy. The proposed methods are particularly well-suited for large-scale sparse data and demonstrate significant improvements over truncated SVD and existing Hadamard decomposition techniques on both synthetic and real-world datasets, confirming their computational efficiency and competitive performance.

0 citationsRead paper

On the GitHub Actions Language: Usage, Evolution, and Workflow Reliability

May 26, 2026

This study addresses the critical issue of frequent failures in GitHub Actions workflows, which severely undermine automation reliability and maintainability. For the first time, it systematically maps 197 language constructs to 14 workflow capability features through a large-scale quantitative analysis of over 260,000 workflows across 49,000 repositories. By integrating language construct categorization with metadata mining, the work uncovers prevalent usage patterns, evolutionary trends, and their impact on workflow reliability. The findings reveal that only a small subset of constructs is heavily used, and that specific capability features are significantly associated with elevated failure rates and maintenance costs. These empirical insights provide actionable guidance for optimizing workflow design and improving robustness in continuous integration and delivery pipelines.

0 citationsRead paper